Sheelah Kolhatkar's cover story today in Bloomberg BusinessWeek about the SEC's hunt to capture Steve Cohen is pretty amazing, and depending on your priors will leave you impressed or infuriated or both with the SEC. I vote both, but I always vote both.
The core of it is the story of how Sanjay Wadhwa, a senior enforcement lawyer at the SEC, got a tip from FBI agent B.J. Kang "that something big might have gone down during the summer of 2008 at SAC Capital," though "It’s not clear whether Kang was motivated by information or intuition." This nebulous tip led Wadhwa to research all previous SEC referrals about SAC. One that he found was a "multipage [September 2008] letter from NYSE Regulation ... [that] said that someone from RBC Capital Markets had pointed out evidence of a market-moving information leak about Elan. 'If there was a leak of information,' the letter read, 'it was probably during the ICAD [International Conference on Alzheimer’s Disease] conference,' when doctors and investors would have been mingling and socializing." Good tip!
This thesis turned out to be incorrect, but the letter did prompt the SEC to launch one of its largest investigations. It would end up issuing 140 subpoenas and amassing 2 million pages of documents as it built a case that kept leading in the direction of SAC Capital. ... They tried to piece together an explanation for the astonishing amount of money SAC had made trading Elan. ... The initial stages involved painstaking work: The firm’s trading records were a jumble of activity with nothing broken out. It was also difficult to discern which of SAC’s 900 employees they should focus on. ...
After sifting for months through every phone call to SAC from anyone connected to Elan, the SEC team pinpointed Martoma and his source, a neurologist and Alzheimer’s expert named Dr. Sidney Gilman, who worked as a consultant to hedge funds through Gerson Lehrman. ... As they tracked Martoma further back in time, a pattern emerged: Over the course of 2007 and 2008, Martoma and Gilman had spoken every time an Elan safety monitoring committee held a meeting.
Eventually they brought the case to prosecutors who arrested Martoma. Given the public information - mostly from the SEC and prosecutors at this point, but still - it's pretty easy to believe that the SEC and prosecutors have Martoma dead to rights; Gilman has told prosecutors that he gave Martoma tons of inside information and Martoma then traded on it. So this really is - apparently - the story of a dogged team of investigators pursuing a thin lead and, through long hours of rigorous detective work, actually catching a criminal. Not Steve Cohen, but someone one level removed from him.
That is impressive. It's hard dogged work; I am depressing myself just thinking about reading phone records for months on end. They get points for, like, the pure arete of it.
But it also sucks, doesn't it? For one thing, I get an unpleasant whiff of data mining from the story. Lots of insider trading cases are built on circumstantial evidence: Insider X at Company Y called Trader Z on the phone, and two minutes later Trader Z bought a lot of Company Y stock, and then the next day Company Y announced it was being acquired. Insider X and Trader Z were social friends, and without a wiretap you don't know what they discussed, but what are the odds that that would happen innocently?
That's a good question but the answer depends on the sample set. If the odds are one in a million, and the SEC looked into twenty million phone calls between Insider X and Trader Z and other insider-and-trader friend pairs like them and brought only one set of charges, then ... that looks a little different, doesn't it?
This is sort of a churlish thing to point out in this case because by all non-circumstantial indications (Gilman's cooperation, etc.) they got it right here, but still. The sheer amount of records reviewed might make you wonder a little about, say, Rajat Gupta, whose insider-trading tipping conviction was based mostly on an awkwardly timed phone-call-and-stock-buying situation. That call looks ridiculous in isolation, but knowing how many millions of other calls the SEC looked at might change your mind.1
This line of thinking might get you worried about Kolhatkar's discussion of "edge":
When a former trader from Galleon Group was asked what the word “edge” meant to him, he laughed and said that from the day he started at Galleon to the day he left, it was probably the most commonly used term around the office. It was such a priority, the trader added, that if you didn’t have it, you’d be quickly left behind. It meant that you knew something that others didn’t. Another trader—a witness in the government’s insider-trading investigation—was asked if he knew of any hedge fund that didn’t traffic in illegal information, according to the person familiar with the inquiry: No, the source answered, they would never survive. In this way, trading on nonpublic material information is similar to doping in professional cycling: Once someone like Lance Armstrong starts doing it, everyone else has to as well. Proving that it’s happening is just as difficult. “These cases take real detective work,” says Tom Sporkin, a partner at Buckley Sandler and a former senior enforcement official at the SEC. “It’s like finding the needle in the haystack.”
This is a deeply troubling passage, isn't it? If all "edge" - everything that makes trading informed and markets efficient - is really insider trading,2 then ... well, then, for one thing, prosecuting insider trading might be really bad for market efficiency. But more important spending months pursuing one insider trading case would be crazy! If this witness's theory is right, then sending Raj Rajaratnam or Mathew Martoma or Steve Cohen or anyone else to jail for insider trading is like a random lightning strike: of the thousands of hedge fund traders who, in this weird view of the universe, are all insider trading, why bother putting a couple of dozen in jail?3 Or, if the theory is wrong, but the SEC treats any edge as evidence of wrongdoing and a reason to start an investigation, then ... then they'll be doing a lot of data-mining on anyone who makes money trading?
The main reason to be worried about this investigation, though, is what it says about priorities. I mean, you don't have to be worried about priorities here: if you believe that insider trading is The Most Important Crime, then catching it is The Most Important Thing We Can Do. I don't believe that, and I feel like a lot of people are starting to doubt it, especially after events of the last few years. Lots of financial misbehaviors were revealed, some of which (Madoff, mortgages, CDOs, whatever) had enormous consequences on the financial system and individuals' life savings, and others of which were insider trading.
If you don't believe that insider trading is the most important financial regulatory problem, then you might worry about the sheer person-hours devoted to this case; the millions of documents and thousands of phone calls that consumed all of several skilled SEC lawyers' time for basically three years. Surely that time and money could be spent elsewhere.
But more than that, I feel like this investigation says something troubling about the skills the SEC develops and prioritizes. Finding needles in haystacks isn't a great skill for a financial regulator. It's okay! I mean, insider trading is a little bad, so someone should try to catch it. And other sorts of actually harmful financial crimes share similar characteristics. Would that someone at some regulator had gone looking for an actual bank statement for Peregrine Financial, for instance.
Others, though, less so. Imagine the sort of financial regulator who, blessed with complete granular information about the London Whale's trading positions - which US financial regulators had4 - would call up Jamie Dimon and say "hey you gotta cut back risk in your whalery, not in a bullshit RWA-and-VaR-model-manipulation way but by actually paring down positions." Or the sort of regulator who, after reading her tenth CDO-squared prospectus, would get a little nervous about home-price-appreciation and correlation assumptions and dig into the modeling and call banks to say "hey what is up with this?"
Two questions you might ask are:
- Would you want that sort of regulator? and
- Is that skill developed by finding the one phone call in a thousand that proves insider trading?
Your answers may vary but mine are "yes" and "no." The paradigmatic SEC investigation - "find an insider trader through phone records" - is about drilling down, not broadening out. It starts from a suggestive general pattern - "boy SAC makes a lot of money" - and looks for the one specific fact to nail somebody. The financial regulators you'd really want would start from specific facts and look for the general pattern. They'd spend years looking for broad problems with systems, not phone records to prove a single instance of wrongdoing by a single person. These SEC lawyers - the ones held up as models of SEC enforcement, the ones responsible for the SEC's one post-crisis success story - should have been finding Bin Laden, not overseeing a financial system.
1.And you can even imagine something similar happening here. Not to Martoma, necessarily, but to Cohen, who ended up going from a large long to a large short in Elan stock after a meeting with Martoma, who was in possession of nonpublic information about bad drug trial results. "What are the odds," you'd ask, "that Cohen and Martoma had a totally innocent meeting and then Cohen dumped all his ELN stock and put on a massive short the next day?" Well, low. One in N, for some large N, right? But if I told you that the SEC looked into 5N conversations that Cohen had just before making other big trades, where does that leave you?
Yeah yeah this is different - his secret conversation was with a guy who had inside information - but nobody knows what was said. If Martoma said "my inside source told me to dump ELN," Stevie's cooked. If he gave a deeply researched fundamental argument that made no evident use of inside information, Stevie's just an innocent victim. Again, "what are the odds?" is the right question, but the answer depends on the sample size.
2.By the way you don't have to believe that - I obviously don't - to notice another thing I harp on, which is that all of the evidence is that some sort of insider-trading-lite is pervasive, so why put such emphasis on imprisoning people who do the slightly more obvious version?
Note here that the SEC "sift[ed] for months through every phone call to SAC from anyone connected to Elan." There were apparently months' worth of calls between Elan insiders and SAC. A few of those calls constituted insider trading, in the SEC's telling. The rest did not. Were they purely social?
3.A more subtle theory would distinguish "edge" - just, like, having an idea or information or insight or model or whatever that you believe to be better than what others have - from "black edge," which is apparently SAC's hilarious term for illegal inside information:
[An SAC trader] wrote that Martoma was acting like he had “black edge.” A trader’s edge is his advantage; it’s the work he’s done and the things he knows about a company. “Black edge,” according to people familiar with the inquiry, is likely a term for information that cannot be doubted and that no one else has. It’s the kind that can make a trader millions and the kind that can put a trader in jail.
And the kind that can get a trader laughed at on Twitter.
4.I'm stretching a little because the Whale is topical; he's really more of the Fed's/OCC's purview than the SEC's. Part of that is because the investment banks who were in the SEC's purview all disappeared due, in part, to inept SEC oversight.